Work-in-progress
This repository will contain the code for the overlapping speech detector described in the paper:
M. Kunešová, M. Hrúz, Z. Zajíc, V. Radová: "Detection of Overlapping Speech for the Purposes of Speaker Diarization", to appear at SPECOM 2019
-
Clean up and add all the missing code:
Added:
- Python code for the overlap detector itself
- MATLAB code for creating artificial overlap data (wavs + reference)
Still Missing:
- Scripts for preparing training data for the OL detector (wav -> h5)
- Scripts for evaluation
-
Resolve external dependencies